Sie sind auf Seite 1von 42
Primary funding is provided by The SPE Foundation through member donations and a contribution from

Primary funding is provided by

The SPE Foundation through member donations and a contribution from Offshore Europe

The Society is grateful to those companies that allow their professionals

to serve as lecturers

Additional support provided by AIME

to serve as lecturers Additional support provided by AIME Society of Petroleum Engineers Distinguished Lecturer

Society of Petroleum Engineers Distinguished Lecturer Program

www.spe.org/dl

1

Comforting, Confusing, Scary: Geomechanical Issues in Naturally- Fractured Reservoirs Prof Gary D Couples Heriot-Watt
Comforting, Confusing, Scary: Geomechanical Issues in Naturally- Fractured Reservoirs Prof Gary D Couples Heriot-Watt

Comforting, Confusing, Scary:

Geomechanical Issues in Naturally-

Fractured Reservoirs

Prof Gary D Couples

Heriot-Watt University, Edinburgh, Scotland

Gary D Couples Heriot-Watt University, Edinburgh, Scotland Society of Petroleum Engineers Distinguished Lecturer

Society of Petroleum Engineers Distinguished Lecturer Program

www.spe.org/dl

With thanks to: Helen Lewis (Hydro-DDA development,

geomech sims), Jingsheng Ma (flow upscaling), Jean-Marie

Questiaux (reservoir models), Mark Reynolds (H-DDA models), Dave Stearns (how to swim upstream)

2

Gary Couples

Geological education (rock mechanics)

Industry employment (Amoco, others)

Academic position Glasgow Uni (hydrogeology)

Moved to Heriot-Watt in 1998

Now, partly engineer, partly geoscientist

Aims of this Talk

Quash some unhelpful myths about fractured

reservoirs

Introduce a wee dose of process understanding (geomechanical interactions)

Outline next-generation approaches to predicting reservoir performance

And comment on what we can do now…

Sequence of Talk

Brief overview of fractured reservoirs

Simple (but incorrect) notions about fractures

Geomechanics of blocky systems, and controls

on effective flow properties

Towards a workflow

All of this applies to “fractured reservoirs”, but

also to unconventional plays where natural fracturing is a major issue

Meaning of Term “Fracture”

A planar feature (resulting from deformation) that disrupts

the continuity of a rock

Here concerned with “open” fractures that locally cause new void space (increase porosity, permeability)

cause new void space (increase porosity, permeability) Shear fracture Joint, extension fracture Compaction band,
Shear fracture
Shear
fracture

Joint, extension fracture

Compaction band, solution seam

Compactional shear band

(granulation

seam)

Open (dilational)

shear fracture

Thanks to Atilla Aydin for expressing some ideas leading to this image

6

Aha! We have a Fractured Reservoir

Map a trap

Aha! We have a Fractured Reservoir Map a trap Wireline data is promising Wireline logs Petrophysical

Wireline data is promising

Wireline logs

Wireline logs

Petrophysical

analysis

data is promising Wireline logs Petrophysical analysis 0 S w 100 Drill the discovery well; it

0

S w

100

Drill the discovery well; it has shows through the reservoir interval

Calculate high oil

saturations

the reservoir interval Calculate high oil saturations 5000 bpd Then the core data arrives Core plug
5000 bpd Then the core data arrives Core plug Estimate flow PI = very small
5000 bpd
Then the core data arrives
Core plug
Estimate flow
PI = very small (1.2 x 10 -4 bbls psi –1 day –1 )
< 6 bpd !
P res = 5 x 10 3 psi
Core plug Estimate flow PI = very small (1.2 x 10 -4 bbls psi –1 day

Service

Company

Report

K = 0.47 mD

Oh, no!

It’s a fractured reservoir

7

Perform well test:

Oh, no! It’s a fractured reservoir 7 Perform well test: Great! We are rich! STOIIP =

Great! We are rich!

STOIIP = GRV x S o x f x NTG x B o

With a large trap… possibly a few billion barrels

“Dark Energy” and “Dark Matter” “Dark Permeability” “ One of the biggest challenges is to
“Dark Energy” and “Dark Matter” “Dark Permeability” “ One of the biggest challenges is to

“Dark Energy” and “Dark Matter”

“Dark Permeability”

One of the biggest challenges is to identify the source of dark permeability, which is inferred to exist because of flow rates that are higher than can be

explained by any arrangement of the known permeability elements

(Richard Steele, BG Group, EAGE Workshop, London, June 2013)

Skewed Production

An example fractured reservoir has the following characteristics:

Super-Giant with ~ 40 years production, ~130 wells: 12% of Cumulative Production from only 1
Super-Giant with ~ 40 years production, ~130 wells:
12% of Cumulative Production from only 1 well
50% of Cumulative Production from only 8 wells
67% of Cumulative Production from only 13 wells
!
Lorentz-type plot % wells Drill these! % flow
Lorentz-type plot
% wells
Drill these!
% flow

If we could reduce development costs, by drilling, say, only 20% of the wells, this would make a major economic

impact…

There is a strong incentive to identify

better development/management

strategies (why do wells “work”?)

9

Data Sources for Fractured

Reservoir Models

Data Sources for Fractured Reservoir Models • Borehole images – Identification of planar features that have

Borehole images

Sources for Fractured Reservoir Models • Borehole images – Identification of planar features that have

Identification of planar features that

have acoustic/electrical contrast

Hall and Lewis (2007)

Seismic anisotropy

contrast Hall and Lewis (2007) • Seismic anisotropy – Estimation of orientations and intensities • Outcrop

Estimation of orientations and intensities

anisotropy – Estimation of orientations and intensities • Outcrop analogues – Observable patterns in a presumed

Outcrop analogues

of orientations and intensities • Outcrop analogues – Observable patterns in a presumed analogue •
of orientations and intensities • Outcrop analogues – Observable patterns in a presumed analogue •

Observable patterns in a presumed analogue

Geomechanical simulation

Questiaux et al (2010)

Distribution of strain, down-scaled to

create fracture distribution

of strain, down-scaled to create fracture distribution – Flow performance data – Well test, history-matched

Flow performance data

Well test, history-matched production

performance data – Well test, history-matched production Note that curvature is not a robust predictor of
performance data – Well test, history-matched production Note that curvature is not a robust predictor of

Note that curvature is not a robust

Causes of (Natural) Fracturing

Fractures are a means of achieving strain need to

understand the strain requirements

Although one can identify a stress criterion for fracture initiation/growth, it is mis-leading (and

limiting) to think of fractures being caused by stress

(and limiting) to think of fractures being caused by stress Example from Olson (2007) showing fracture

Example from Olson (2007) showing fracture patterns that form in two progressively-evolved models driven by strain boundary conditions

models driven by strain boundary conditions Fold-associated fracture systems, and mechanical-unit

Fold-associated fracture systems, and mechanical-unit boundaries, described by Stearns (1968), Lewis and Couples (1992)

11

Reservoir Models

Using data, and adopting a conceptual model

Discrete Fracture Network, and/or

Fracture corridor “objects”

Match to fracture intensity in vertical wells
Match to fracture intensity
in vertical wells
“objects” Match to fracture intensity in vertical wells DFN model based on wellbore image data +

DFN model based on wellbore image data + seismic and fitted to a fracture corridor concept (from PhD work by Salah al Dhahab)

Sector model of a fractured carbonate reservoir with fracture corridors (after Questiaux et al 2010)

12

Pressure-Permeability Coupling

This is the main point!

Many low-permeability reservoirs are affected by

fractures

Flow performance suggests that the effective permeability often changes during production

This coupling explanation:

[k frax = f (P, s)]

requires an

Change in fracture connectivity?

Change in fracture apertures?

in fracture connectivity? – Change in fracture apertures? Geomechanics • The simple causation models usually

Geomechanics

The simple causation models usually adopted,

calibrated to production history, used to predict performance, used as rules-of-thumb are wrong

The Fundamental Flaw

The existing, simple rules are based on

mechanical ideas derived from elastic, free

bodies whose circumstances cannot be applied to real-world systems

The simple mechanical models that underlie

the rules assume a constant stress state

during movements of the fracture walls

This cannot be true in Nature

Ideas Examined Here

Fracture opening/closing = changes in stress

So classical arguments about fracture apertures

being controlled by effective stress are wrong

Fracture-parallel stress and fracture opening

Effective stress and poro-elasticity

Stress heterogeneities in blocky systems

A Simple Truth

D fracture aperture = rock movement

Surrounding blocks, so aperture increases can only occur by shortening the adjacent rock matrix (which will increase its stress), or by

lengthening the whole mass and the reverse

is also true for aperture closure

mass – and the reverse is also true for aperture closure fracture y d = 1
fracture y d = 1 m 16
fracture
y
d = 1 m
16

x

This is based on treating the rock as an elastic material

for aperture closure fracture y d = 1 m 16 x This is based on treating
for aperture closure fracture y d = 1 m 16 x This is based on treating
for aperture closure fracture y d = 1 m 16 x This is based on treating
for aperture closure fracture y d = 1 m 16 x This is based on treating
for aperture closure fracture y d = 1 m 16 x This is based on treating
for aperture closure fracture y d = 1 m 16 x This is based on treating

Adjacent blocks

An Analogy

An easy thought-experiment

An Analogy An easy thought-experiment The car jack represents the pore pressure opening the fracture apertures

The car jack represents the pore pressure opening the fracture apertures

In the lecture room,

the chairs are arranged

side-by-side

Now, assume that the space between them is

increased BUT, the

length of the row is not changed…

This is only possible if the size of each chair is reduced

This “Problem” is Already Known

In a hydrofrac well stimulation, a similar behaviour occurs,

when fractures open and hence load the sideburden. Stress

changes caused by one stage of treatment interfere with the next (adjacent) stages, and there is uplift of the ground and tilting (we use tilt-meters to monitor this).

The same phenomenon is observed in geotechnical situations

such as the placement of concrete diaphragm walls

The opening

as the placement of concrete diaphragm walls The opening Soil is then removed from here Uplift

Soil is then removed from here

Uplift & tilting stress change
Uplift & tilting
stress change

Stress is not constant

Fractures Parallel to Load

Simple model of elastic solid with elliptical opening

Model is loaded at boundaries, and the aperture changes

If s y > s x , the model itself gets wider!

Oops… same

problem…. and s x would have to increase… s y a b s x s
problem…. and s x would have to increase…
s
y
a
b
s x
s x
Note: this is the model used to calculate
wellbore stability, and it is wrong
s
y

19

Poro-Elasticity

Poro-Elasticity High fluid pressure causes the elements of the rock framework to shrink, leading to fracture
Poro-Elasticity High fluid pressure causes the elements of the rock framework to shrink, leading to fracture

High fluid pressure causes the elements of the rock framework to shrink, leading to fracture opening (usually)

In poro-elastic terms, an extra stress acting

parallel to the fracture causes the fracture to

close…Oops

Strain normal to fracture (=P/E)

Simple lattice model after Couples (2014)

When stress component parallel to fracture is 3x Pp increase, fracture closes (and the reverse occurs, too)

1.50E-04 1.00E-04 Case 1 Case 3 5.00E-05 Case 4 Case5 Case 6 0.00E+00 -5.00E-05 0.00
1.50E-04
1.00E-04
Case 1
Case 3
5.00E-05
Case 4
Case5
Case 6
0.00E+00
-5.00E-05
0.00
0.10
0.20
0.30
0.40
0.50
closes
opens

Poisson’s ratio

Simple elasticity is not sufficient

0.00 0.10 0.20 0.30 0.40 0.50 closes opens Poisson’s ratio Simple elasticity is not sufficient 20

20

Why?

These “difficulties” with models occur because

geologists (and others) have been taught to

think of stress as fixed, or at least arbitrary (and we are seduced to use elasticity because

of its simplicity)

A systems approach to Geomaterials highlights the fallacy of that viewpoint

Indeed, stress is the dependent parameter

that indicates the mechanical state

Stress Heterogeneity

“No fracture knows about the far-field stress” quote

attributed (in First Break) to Gary Couples at EAGE Workshop in London, June 2013

Two examples of stress state across a system

involving discontinuities

Far-field loading is simple and uniform, but evolved internal state is not

is simple and uniform, but evolved internal state is not Hall et al. (2007) Black lines

Hall et al. (2007)

Black lines show orientation and magnitude of s 1 Grey lines show block boundaries

and magnitude of s 1 Grey lines show block boundaries Baghbanan and Jing (2008) If new

Baghbanan and Jing (2008)

If new fractures are created aligned with current s 1 , then they would not be parallel to existing fractures, so we would not expect good orientation statistics

22

What Is Going On?

Blocky geomechanics results in non-linear,

interactions (and these are NOT elastic)

Stress state is not homogeneous

Pore fluids provide another, bi-directional

interaction

Thermal effects add a third interaction axis

System response is not deterministic, with emergent behaviours

Understanding the responses requires use of

numerical simulations

23

Hydro-DDA

Examples to follow are based on the 2D

simulation environment Hydro-DDA

This couples single-phase fluid flow with a discontinuum simulator that deals with the geomechanics of fractured/blocky systems

DDA stands for Discontinuous Deformation

Analysis (Lin, 1995; Shi and Goodman, 1998)

Hydro-DDA was created by Helen Lewis and Mo Rouainia (Rouainia et al 2006)

24

Flow Simulation Results Note that the fractures with most flow are not easily predicted by

Flow Simulation Results

Note that the fractures with most flow are not easily predicted by knowing the shortening direction

are not easily predicted by knowing the shortening direction Colours: pressure (head) contours White arrows: Darcy

Colours: pressure (head) contours

White arrows: Darcy velocity

pressure (head) contours White arrows: Darcy velocity Loading arrangements These results DO NOT support the idea

Loading

arrangements

These results DO NOT support the idea that the resolved effective stress governs fracture aperture/conductivity

Predicting apertures (and hence flow) from fluid pressure alone, or loading, is not a sensible thing to do

25

Similar analysis with uniform fracture patterns

Colours show pressure distribution Numbers are perm ratio

A Regular Fracture Pattern

100x 10x Reynolds, 2004; Reynolds et al 2007 250x 2.5x 26
100x
10x
Reynolds, 2004;
Reynolds et al 2007
250x
2.5x
26

Same model (symmetric, regular spacing) in each panel, but different loadings (as in previous slide)

Note the variability of flow pattern, which translates to different effective perms in

every case

Highest eff perms are NOT in cases where current load is same as causative load!

In a reservoir with non-uniform stress state (the norm), identical fracture patterns have different effective properties that depend on the local conditions

Log 10 eff perm (mD)

Parameter Combinations

In quite a few model

configurations, we see

a major effective- perm enhancement related to a small

change in model

parameters often over a limited range

In non-linear terms,

this represents a

bifurcation behaviour

Interestingly, the “odd” results

seem to occur mostly when there

is a strong fluid energy gradient

3-orders of magnitude change!

4

3

2

1

fluid energy gradient 3-orders of magnitude change! 4 3 2 1 Each curve derived from one
Each curve derived from one model
Each curve derived
from one model

Loading parameter

You can imagine the challenge of finding all of these in the multi- dimensional parameter spaces

27

Tiny Changes in Connectivity

fracture) can have a major impact 2 cm
fracture) can have a major impact
2 cm

This comparison shows how a dynamic change (local opening or closure of a single

The two models have “identical”

fracture network statistics

Colours show pressures. Black and white lines are fracture traces. Difference between models

is very small only a tiny change in connectivity of the fracture network (see inset box). The

effective perm of the right model is 6X that of the left one. This change cannot be predicted from fracture population statistics.

Keff y (mD)

Keff x (mD)

Geomechanics+Fluids+Thermal

Adding another interaction possibility

Particularly relevant for injection wells

But also applicable in many enhanced recovery processes (thermal methods)

Model at right shows simulation to calculate effective perms of a fractured geothermal

rock mass, under load, stimulated by cold-

water injection and then allowed to re- equilibrate. Permanent improvements in

perms range from about 50x to 10x, depending on distance from injection well.

The process involves block movements that lead to propping of fracture openings

Model and loading

Simulation outcome Fluid Flow
Simulation outcome
Fluid Flow
openings Model and loading Simulation outcome Fluid Flow 350 330 310 290 270 250 230 210

350

330

310

290

270

250

230

210

190

1800

1600

1400

1200

1000

800

600

400

200

K xeff vs position

After cooling

0 20 40 60 80 100
0
20
40
60
80
100

Distance from wellbore (ft)

9hrs 5 hrs 1 hr After re-heating 100 0 0 20 40 60 80
9hrs
5 hrs
1 hr
After re-heating
100
0
0
20
40
60
80

Flores et al (1995)

29

A Real One…

Injectivity test, at three rates (5000, 10000, 15000 BWPD)

Big increase in eff perm after first flow period

Operator says this is NOT due to induced fracture (P inj well below s frac )

Do you have any other real examples to share?

r a c ) Do you have any other real examples to share? thanks to Peter

thanks to Peter Roberts

Could be an example of the effects just described

Effective Flow Properties

Highly heterogeneous aperture distributions

and variable fracture network characteristics

Flow effects depend on aspects of the system that are not fixed, but which change as a function of the global and local conditions

Changes often do not follow a simple path, so

simple-idea rules not adequate but perhaps

some functional relationships can be defined

Static analysis cannot capture these effects

Some Application Areas

“Normal” fractured reservoirs

Unconventionals stimulation treatments

interact with natural fracture systems

Thermal recovery strategies

Combined heat and fluid loads

Can we engineer new flow paths?

“Conventional” reservoirs – injection issues

Drilling through fractured systems…

new flow paths? • “Conventional” reservoirs – injection issues • Drilling through fractured systems… 32

32

Real-World Applications

The significant non-linearities, and thus

unexpected jumps in effective properties, are

often associated with high gradients

Consider where/when/why such gradients may occur in your reservoir

1500m
1500m

Questiaux et al (2010)

0.90 diff relperms S o 0.35 33 same relperms
0.90 diff relperms
S o
0.35
33
same relperms

Gradients due to:

Fracture corridors Proximity to wells

Fluid saturation boundaries

Thermal changes Structure reactivation

These changes will have a big impact on upscaled relperms in frax corridors

So, Moving Towards a Workflow…

Now feasible to run reservoir simulations that

include geomechanics (various levels of

fidelity and realism)

Due to coarse cells, these HAVE to use pseudo-functions to describe responses

To capture the uncertainties, need multiple

stochastic runs

We still need to do more work to calculate the full range of pseudo-functions

Fracturing = Strain = Flow Props

Stratification of lithologies

leads to mechanical layering

Most natural deformations

exploit the layering (flexural

slip processes)

So, fractures will occur in characteristic patterns that allow the strains to develop We can exploit that knowledge!

There is a new project starting on this…

frax pattern >> eff flow props >> simulation

(a) Example fracture patterns

props >> simulation (a) Example fracture patterns (b) (c) e y Model 17b e x (d)

(b)

>> simulation (a) Example fracture patterns (b) (c) e y Model 17b e x (d) k

(c)

>> simulation (a) Example fracture patterns (b) (c) e y Model 17b e x (d) k

e y

Model 17b
Model 17b

e x

Example fracture patterns (b) (c) e y Model 17b e x (d) k x Plastic strain
Example fracture patterns (b) (c) e y Model 17b e x (d) k x Plastic strain
(d) k x
(d)
k
x

Plastic strain e x

(c) e y Model 17b e x (d) k x Plastic strain e x e x

e x

Load step 5 Load step 45
Load step 5
Load step 45

But, caution is needed:

These are static effective perms…

But, caution is needed: These are static effective perms… k x = 10 k y =
k x = 10
k x = 10

k y = 44

(referenced to matrix k)

perms… k x = 10 k y = 44 (referenced to matrix k) k x =
k x = 54
k x = 54

k y = 45

(referenced to matrix k)

Two models with same strain, but different flow properties

35

What to Do Now??

Best Practice:

Assess the likely current mechanical state (heterogeneous) across the reservoir (including how it

developed over geological time)

Simulate coupled flow/mechanics models that capture the local situations leading to effective properties for

input into cells

Calibrate these with well-test data

Run full-field simulations (coupled, if possible)

Assess the potential impact of extreme excursions in

effective properties

Comforting, Confusing, Scary

Parts of some reservoirs have fracture

distributions that interact with the actual

fluid/thermal/mechanical states in simple ways

it may be possible to deal with these circumstances using pseudo-static reservoir

models

comforting

In other cases, strongly non-linear interactions

can be expressed via major changes in effective

properties, and we need coupled models (or info

derived from them

)

confusing , or scary

Sorry to be so pedantic!

A Final Word About Stress

It is a measure of the STATE of a material/system

It is thus an intensive parameter, and so it is not

conserved and cannot be moved around (as can extensive quantities)

In equations, stress is the dependent variable, related to strain via the stiffness (the problem with elasticity is the linearity of the eqns)

In the absence of changes in stiffness, stress

cannot be constant when strain occurs

Ideas based on the notion of constant stress are not valid when rocks change shape

Myth-Busting

The rules-of-thumb that are in

common use are not based on

well-argued analysis

They have become myths

I have decided to take a risk and

ask some simple questions

Outcomes challenge the

predictions

simple questions • Outcomes challenge the predictions www.4thmedia.org • As a consequence, the Emperor’s

www.4thmedia.org

As a consequence, the Emperor’s backside looks unattractive…

Summary

Remember: stress does not move, although the distribution of stress may change

In a fractured reservoir, it is not only the fluids that

move! (the rocks move a lot!)

Geomechanics is involved all the way from reservoir creation to abandonment, with particularly important expression during the production phase

Avoid the pitfall of “pretending” to do geomechanics

by making a few calculations based on the wrong assumption that stress is constant

Expertise is available to help but you want to be aware of the value (and cost) of making things more complicated

End

I wish to express my thanks to colleagues, students, questioners, and critics who, over many years, have challenged me to continually develop my ideas on fractures and geomechanics

References:

Baghbanan, A. and Jing, L. 2008. Stress effects on permeability in a fractured rock mass with correlated fracture length and aperture. International Journal of Rock Mechanics and Mining Sciences, 45, 1320-1334. Couples, G.D. 2014. Geomechanical impacts on flow in fractured reservoirs, In. Spence, G.H., Redfern, J., Aguilera, R., Bevan, T.G., Cosgrove, J.W., Couples, G.D.& Daniel, J.-M. (eds) Advances in the Study of Fractured Reservoirs. Geological Society, London, Special Publications, 374, http://dx.doi.org/10.1144/SP374.17 Flores, M. Davies, D., Couples, G. and Pallson, B. 2005. Stimulation of geothermal wells: can we afford it? Proceedings World Geothermal Congress 2005, Ankara, Turkey. Hall, S.A., Lewis, H. and Macle, X. 2007. Improved seismic identification of inter-fault damage via a linked geomechanics seismic

approach, . In: Lewis, H. and Couples, G.D. (eds) Relationships Between Damage and Localization, Geological Society of London,

Special Publications, 289, 187-207. Lin, C.T. 1995. Extensions of the discontinuous deformation analysis for jointed rock masses and other blocky systems. PhD thesis, University of Colorado.

Questiaux, J.-M., Couples, G.D., Ruby, N. 2010. Fractured reservoirs with fracture corridors. Geophysical Prospecting, 58, 279-295, doi: 10.1111/j.1365-2478.2009.00810.x Reynolds, M.A. 2004. Load-sensitive fluid flow through fracture-matrix systems. PhD thesis, Heriot-Watt University, 239p. Reynolds, M.A., Couples, G.D., Lewis, H. and Pickup, G.E. 2007. Localization processes in a coupled hydrogeomechanically-sensitive fractured system. In: Lewis, H. and Couples, G.D. (eds) Relationships Between Damage and Localization, Geological Society of

London, Special Publications 289, 209-225.

Rouainia, M., Lewis, H., Pearce, C., Bicanic, N., Couples, G.D., and Reynolds, M.A. 2006. Hydro-geomechanical modelling of seal behaviour in overpressured basins using discontinuous deformation analysis. Engineering Geology 82 222-233. Shi, G.H. and Goodman, R. 1998. Generalization of two-dimensional discontinuous deformation analysis. International Journal for Numerical and Analytical Methods in Geomechanics, 13, 359-380.

Your Feedback is Important Enter your section in the DL Evaluation Contest by completing the

Your Feedback is Important

Enter your section in the DL Evaluation Contest by completing the evaluation form for this presentation

http://www.spe.org/dl/

evaluation form for this presentation http://www.spe.org/dl/ Society of Petroleum Engineers Distinguished Lecturer

Society of Petroleum Engineers Distinguished Lecturer Program

www.spe.org/dl

42
42